Complex chromatic imaging for enhanced radar face recognition

Abstract

Face recognition with millimeter-wave radar surpasses traditional cameras with better range, less intrusion, and safe material penetration using non-ionizing radiation. However, using complex-valued millimeter wave radar data for face recognition encounters challenges in extracting and representing features due to its complex nature and compatibility issues with high-performing image-based recognition systems. This paper introduces a novel approach utilizing Complex Chromatic Images (CCI) to address these challenges and enhance radar-based face recognition. Proposed Complex Chromatic Images retain both the magnitude and phase information of radar signals, providing a comprehensive representation of facial characteristics. A Complex Chromatic Image-Convolutional Neural Network (CCI-CNN) is developed to extract features from Complex Chromatic Images. Various sub-space analysis techniques are employed to tackle the high-dimensional nature of the complex-valued data. The effectiveness of the proposed approach is evaluated using various classifiers like Support Vector Machine (SVM), Random Forest (RF), and Convolutional Neural Network (CNN). Extensive experimental results and different evaluation metrics reveal that the proposed images approach consistently outperforms the conventional complex data images. Furthermore, when compared to existing mm-wave radar face recognition methods, our approach stands out with an impressive 99.7% accuracy. This study showcases superior recognition performance on complex-valued data, successfully addressing a large multiclass scenario with 206 distinct classes.